Ask HN: How to Improve AI Usage for Programming
Analysis
The article describes a developer's experience using AI (specifically Claude Code) to assist in rewriting a legacy web application from jQuery/Django to SvelteKit. The author is struggling to get the AI to produce code of sufficient quality, finding that the AI-generated code is not close enough to their own hand-written code in terms of idiomatic style and maintainability. The core problem is the AI's inability to produce code that requires minimal manual review, which would significantly speed up the development process. The project involves UI template translation, semantic HTML implementation, and logic refactoring, all of which require a deep understanding of the target framework (SvelteKit) and the principles of clean code. The author's current workflow involves manual translation and component creation, which is time-consuming.
Key Takeaways
- •The author is seeking advice on how to improve their use of AI for programming, specifically for code translation and refactoring.
- •The primary challenge is the AI's inability to generate code that meets the author's quality standards and requires minimal manual review.
- •The project involves complex tasks like UI template conversion, semantic HTML implementation, and logic refactoring, highlighting the need for AI to understand context and idiomatic style.
- •The author is using Claude Code, but the results are not satisfactory.
- •The goal is to reduce the time spent on manual code review and improve development speed.
“I've failed to use it effectively... Simple prompting just isn't able to get AI's code quality within 90% of what I'd write by hand.”